Assurant's Global Housing segment grew adjusted EBITDA 18% to $287 million ex-catastrophes in the second quarter of 2026, while absorbing $12 million less favorable prior-year reserve development than the year-earlier quarter and guiding to a $71 million full-year reserve headwind against 2025.

Margin expansion alongside a shrinking reserve tailwind is the reverse of what most specialty carriers printed this season.

Key Takeaways

  • The reserve headwind is worth 4 to 5 points of EBITDA growth. Underlying growth is roughly 10% once the $71 million full-year reversal is stripped out, against the mid-single-digit rate management is guiding to including it.
  • Only $42 million of that $71 million had landed by midyear, so management expects the drag to widen through the second half.
  • Premium tracks a servicing metric, not a rate filing. Gross written premium is the placement rate applied to tracked loans, and the sequential fall to 2.02% traces to a client moving part of its portfolio to another servicer.
  • Tracked loans grew 9% to more than 34 million, with a new Freedom Mortgage relationship adding roughly 2.6 million loans ramping over the next 12 months.
  • The 35% non-catastrophe loss ratio still carries weather. It excludes only events above a reportable threshold, leaving sub-threshold convective storm, wind and hail losses inside the figure.

Decomposing an 18% That Came From Four Levers

Company-wide, GAAP net income rose 27% to $298.6 million and adjusted EBITDA excluding catastrophes climbed 18% to $491.4 million, with adjusted EPS ex-catastrophes at $6.60, up 19%. Global Housing's ex-cat adjusted EBITDA of $287 million, also up 18%, rested on a non-catastrophe loss ratio near 35% and lower catastrophe reinsurance costs. Including $12.2 million of reportable catastrophe losses, down from $29.8 million a year earlier, segment adjusted EBITDA rose 28% to $274.8 million.

The reserve line moved the other way. The quarter carried $12 million less favorable prior-year development than Q2 2025, and full-year guidance embeds $71 million less than 2025 companywide, of which $42 million had already been realized in the first half (earnings call transcript, August 2026).

MetricQ2 2026Q2 2025
Global Housing adjusted EBITDA (ex-cat)$287.0M~$243M (implied, +18%)
Global Housing adjusted EBITDA (incl. cat)$274.8M~$215M (implied, +28%)
Non-cat loss ratio~35%Comparable, per management
Catastrophe losses, Global Housing$12.2M$29.8M
Lender-placed placement rate2.02%2.02% (flat YoY)
Tracked loans34M+~31M (implied, +9%)

Strip the reserve reversal out and management guided to roughly 10% underlying adjusted EBITDA and EPS growth, well above the mid-single-digit rate it is actually guiding to. The gap between reported and underlying growth is almost entirely the reserve line, roughly 4 to 5 points of EBITDA surrendered to a shrinking cushion in one year.

The mechanical reading is that current accident-year picks are no longer carrying the margin they carried in 2024 and 2025. A carrier setting picks with margin sees them develop favorably in a benign loss year, which generates exactly the releases Assurant realized then. As redundancy is consumed, favorable development shrinks whether or not underwriting quality changed. Management's own framing, underlying growth near 10%, says the current book prices and reserves adequately without leaning on that cushion.

The Premium Base Is a Servicing Decision

Lender-placed insurance is coverage a mortgage servicer buys when a borrower's own policy lapses or cannot be verified, charged back to escrow. A voluntary book prices to a rate filing against a stable policy count. A lender-placed book's premium is the placement rate, the share of a servicer's tracked portfolio that falls out of voluntary coverage, applied to total tracked loans.

That makes the leading indicator sensitive to single-client decisions. The sequential decline to 2.02% traces to a disclosed cause: CFO Keith Meier attributed it to "a client transferred a portion of their loan portfolio to another loan servicer," removing loans that had carried above-average placement. CEO Keith Demmings described the underlying trend as "very stable sequentially" once that transfer is excluded. Nothing in the number says borrowers are maintaining voluntary coverage better.

The mechanics run both ways. Losing a servicing client shrinks numerator and denominator together when the transferred loans placed above average. Winning one expands the denominator immediately and the numerator only as loans season into tracking and lapses surface. Tracked loans grew 9% to more than 34 million, with Freedom Mortgage, a top-10 servicer, adding roughly 2.6 million loans over the next 12 months. A rising placement rate can therefore mean borrower stress, or it can mean a carrier just won a servicer whose book runs hot.

Loss emergence differs too. Policies are written through the servicer's tracking system with a documented effective date tied to the lapse, which shortens the reporting lag against an agency-sold book. The exposure mix runs the other way, skewing toward distressed, transferred or post-foreclosure properties. The NAIC documents the pricing consequence: because the servicer picks the carrier and the borrower pays, price competition breaks down, and New York regulators found "high prices and low loss ratios" industry-wide. That drove the Real Property Lender-Placed Insurance Model Act in 2020, following a 2013 NYDFS settlement with Assurant carrying a $14 million penalty and restitution.

What "Non-Catastrophe" Leaves In

The 35% figure is not a weather-free loss ratio. It excludes reportable catastrophes, individual events above a set pre-tax threshold net of reinsurance and client profit-sharing, which Assurant discloses separately (Reinsurance News).

Everything below that threshold stays in. Severe convective storm claims that never reach the reporting bar, chronic wind and hail activity, roof and water damage from unnamed events: all of it sits inside the non-catastrophe ratio. The $12.2 million of reportable catastrophe losses is the broken-out tail, not the segment's full weather sensitivity.

That matters for how much of the 18% is durable. Three of the four levers behind it, a lighter reportable catastrophe quarter, lower reinsurance cost, and favorable non-catastrophe experience that itself contains weather, can reverse inside a single bad accident season. The fourth, reserve development, has already turned against the segment while the other three were running for it.

Assurant is further along that path than most. Reserve-release reliance is the standing question across the softening P&C market, and NCCI found the same redundancy erosion in workers compensation as that cycle turned. What distinguishes this print is that the drawdown is visible and quantified rather than inferred, and that the guided $71 million headwind was only $42 million realized at midyear.

Further Reading

Sources